Journal of Insect Science
RESEARCH
Spatial and Temporal Dynamics of Stink Bugs in Southeastern Farmscapes Grant L. Pilkay,1,2,3 Francis P. F. Reay-Jones,1 Michael D. Toews,4 Jeremy K. Greene,5 and William C. Bridges6 1
Department of Agricultural and Environmental Sciences, Pee Dee Research and Education Center, Clemson University, 2200 Pocket Rd., Florence, SC 29506 Present address: Fayetteville Technical Community College, Biology Department, 2201 Hull Rd. Fayetteville, NC 28303 Corresponding author, e-mail:
[email protected] 4 University of Georgia, Department of Entomology, 2360 Rainwater Rd., Tifton, GA 31794 5 Department of Agricultural and Environmental Sciences, Edisto Research and Education Center, Clemson University, 64 Research Rd., Blackville, SC 29817 6 Department of Mathematical Sciences, Clemson University, 101 Sikes Ave., Clemson, SC 29634 2 3
Subject Editor: Wade Worthen J. Insect Sci. 15(23): 2015; DOI: 10.1093/jisesa/iev006
ABSTRACT. A 3-yr study (2009–2011) was conducted to examine the spatial and temporal dynamics of stink bugs in three commercial farmscapes. Study locations were replicated in South Carolina and Georgia, in an agriculturally diverse region known as the southeastern coastal plain. Crops included wheat, Triticum aestivum (L.), corn, Zea mays (L.), soybean, Glycine max (L.), cotton, Gossypium hirsutum (L.), and peanut, Arachis hypogaea (L.). Farmscapes were sampled weekly using whole-plant examinations for corn, with all other crops sampled using sweep nets. The predominant pest species of phytophagous stink bugs were the brown stink bug, Euschistus servus (Say), the green stink bug, Chinavia hilaris (Say), and the southern green stink bug, Nezara viridula (L.). Chi-square tests indicated a departure from a normal distribution in 77% of analyses of the variance to mean ratio, with 37% of slopes of Taylor’s power law and 30% of coefficient b of Iwao’s patchiness regression significantly greater than one, indicating aggregated distributions. Spatial Analyses by Distance IndicEs (SADIE) indicated aggregated patterns of stink bugs in 18% of year-end totals and 42% of weekly counts, with 80% of adults and nymphs positively associated using the SADIE association tool. Maximum stink bug densities in each crop occurred when the plants were producing fruit. Stink bugs exhibited greater densities in crops adjacent to soybean in Barnwell and Lee Counties compared with crops adjacent to corn or fallow areas. The diversity of crops and relatively small size of fields in the Southeast leads to colonization of patches within a farmscape. The ecological and management implications of the spatial and temporal distribution of stink bugs within farmscapes are discussed. Key Words: sampling, Taylor’s power law, patchiness regression, inverse distance weighted, SADIE
The widespread adoption of transgenic cultivars of cotton, Gossypium hirsutum (L.), expressing Bacillus thuringiensis (Bt) toxins to control the heliothine complex and the eradication of the boll weevil, Anthonomus grandis grandis Boheman, have decreased the need for the application of broad-spectrum insecticides on cotton in the southeastern United States (Greene et al. 1999, Bundy and McPherson 2000). This reduction in pesticide use has allowed stink bugs to greatly expand their damage on cotton (Greene et al. 1999, 2001). As Bt cultivars have become more widespread in Asia, South America, and the United States, documented stink bug damage on cotton has increased (Greene et al. 1999, Panizzi and Schaefer 2000, Zeng et al. 2009). Crop losses in U.S. cotton caused by stink bugs were estimated at $31 million in 2008 (Williams 2009). Significant yield losses from this pest complex are also frequent in soybean, Glycine max (L.), with up to $60 million in losses annually in the United States (McPherson and McPherson 2000). Stink bugs also can be serious pests in corn, Zea mays (L.) (Negron and Riley 1987, Ni et al. 2010). Phytophagous stink bugs extract fluids from plant tissues with piercing and sucking mouthparts (McPherson and McPherson 2000). Crops can be damaged by the mechanical and chemical actions of stink bug feeding, resulting in a loss of turgor pressure and injection of digestive enzymes. Pathogens, introduced as opportunistic infections or by direct transmission during feeding, also contribute to cotton losses (Ragsdale et al. 1979, Barbour et al. 1990, Medrano et al. 2007). Feeding damage on a developing cotton boll ranges from stained lint and damaged seeds to pathogen-induced boll rot or boll abortion (Ragsdale et al. 1979, Barbour et al. 1990, Medrano et al. 2007). Stink bugs also transmit yeast-spot disease in soybean (Daugherty 1967). In grain and legume crops, stink bug feeding will decrease kernels or bean quality, and entire
heads or fruiting bodies can be lost (Hall and Teetes 1982, Espino and Way 2008). Depending on growth stage, stink bug feeding on corn can cause low kernel weights, loss of kernel yield, and abortion of small ears (Ni et al. 2010). Edible plant parts may become distasteful as a result of stink bug feeding, with a bitter taste or pithy texture (Callahan et al. 1960). Alternative management strategies must be developed to reduce yield loss and the use of broad-spectrum insecticides currently applied. The predominant pest species of phytophagous stink bugs in the southeastern coastal plain are the green stink bug, Chinavia hilaris (Say), the southern green stink bug, Nezara viridula (L.), and the brown stink bug, Euschistus servus (Say). Stink bugs are highly polyphagous and move between adjacent agricultural and wild hosts in the farmscapes; this movement is linked to crop phenology and the availability of suitable food sources (Jones and Sullivan 1982). Southeastern farmscapes are typically characterized by a mosaic of relatively small fields ( 1), random ( ¼ 1), or uniform distribution ( < 1). The null hypothesis of spatial randomness was rejected for P < 0.025 (aggregation) or P > 0.975 (uniformity). The SADIE association tool was used to determine spatial associations between adults and nymphs for each species and for total counts of all species by farmscape and year. An overall index of association (X) was determined between each paired dataset, with positive associations for X > 0 (P < 0.025) or negative associations for X < 0 (P > 0.975). Mean X was determined from the local spatial associations (Xk) for each sampling point k. A positive association between two variables indicates a patch or gap for both variables, whereas a negative association indicates a patch of one variable and a gap of another (Perry 1997, 1998). Selected SADIE local aggregation indices were imported into the GIS software (ArcView 9.2, ESRI 2006) and the Inverse Distance Weighting (IDW) spatial statistical method was used to visualize stink bug aggregations. Cell values in IDW are interpolated using a linear weighted combination of data points around each cell. SADIE was chosen over more traditional geostatistical methods, such as kriging, because SADIE can illustrate local variability in spatial distribution and association among datasets sharing the same sampling points (Perry et al. 2002). The influence of distance from the field edge and the effect of adjacent crop plantings on stink bug densities were analyzed separately for each farmscape. The response variables were the total numbers of each primary pest species and all species combined for each life stage averaged across sample dates. As the authors did not dictate crop plantings in the commercial farmscapes, not all crop and adjacent crop combinations occurred in all farmscapes. As such, crop and adjacent combinations were combined into a fixed “crop and adjacent crop” effect. “Crop and adjacent crop” effects in Lee County consisted of adjacent fields of corn and cotton, corn and wheat-double-crop soybean, cotton and soybean, cotton and woods, wheat and woods, corn and woods, wheat and cotton, and soybean and woods. Adjacent fields in Barnwell County consisted of cotton and corn, cotton and fallow,
3
cotton and soybean, corn and fallow, soybean and fallow, and soybean/ peanut. In Tift County, where transects were separated by grass borders, adjacent fields consisted of cotton and pines, cotton and pecan, cotton and soybean, cotton and sorghum, and cotton and watermelon. These grass borders were considered to be adjacent to both fields on each side and were combined into a single value. Because only certain fields were used in certain years, an effect combining the two into “field and year” was created. The treatment arrangement of the study was a twofactor factorial of distance from edge and “crop and adjacent crop” combinations. The experimental design of the study was a split plot with subsampling. The whole plot factor was “crop and adjacent crop” arranged in a completely randomized design with “field and year” as replicates. The subplot factor was distance arranged in a randomized complete block design with “field and year” as blocks. The two to three transects within each field were subsamples. A linear model was developed to account for distance, “crop and adjacent crop,” and their interaction as fixed effects, and “year and field” within “crop and adjacent crop” (i.e., whole plot error or errorA), interaction of distance with “year and field” (i.e., subplot error or errorB), and residual error (i.e., subsampling error or errorC) as random effects: Yijkl ¼ u þ FYi þ Cj þ FYðCÞij þ Dk þ C Dik þ D FYðCÞijk þ TðD FYðCÞÞijkl where Yijkl is the response variable in “year and field” i, “crop and adjacent crop” j, distance k, and transect l; u is the overall mean of the response; FYi is the effect of “year and field” i; Cj is the effect of crop j; FY(C)ij is the effect of “year and field” i within “crop and adjacent crop” j (errorA); Dk is the effect of distance k; C*D*ik is the interaction effect of “crop and adjacent crop” j and distance k; D*FY(C)ijk is the interaction effect of distance k and “year and field” i within “crop and adjacent crop” j (errorB); and T(D*FY(C))ijkl is the effect of transect l within distance k and “year and field” i within “crop and adjacent crop” j (errorC). PROC GLIMMIX (SAS Institute 2008) was chosen to estimate and test model terms, as the experimental design of the study involved multiple random effects, resulting in a split-plot design that required correction for the random effects and appropriate error terms for the level of the split plot. Examination of the count data using Proc FREQ determined that the data followed a normal distribution; therefore, no transformations or link functions in GLIMMIX were necessary. Stink bug counts for corn and woods and wheat and cotton adjacent crops were insufficient for analysis in Lee County and were omitted from the model. Significance for model terms was determined using a probability level of 95% (P < 0.05). Degrees of freedom were calculated using the Kenward–Roger degrees of freedom approximation (Kenward and Roger 1997). Because distance is a continuous variable and crop and adjacent crop combinations were considered a single effect, contrast statements were used to evaluate the impact of distance and crop and adjacent effects over traditional pairwise comparisons. As different treatment combinations were present in each farmscape, contrast coefficients were manually assigned as needed.
Results Species Composition and Temporal Dynamics. Across all years and farmscapes, E. servus was the most abundant stink bug (64.7% of all individuals), with fewer N. viridula (12.9%) and C. hilaris (11.0%) counts. Densities of adults and nymphs varied with year and farmscape (Table 1). The rice stink bug, Oebalus pugnax (F.), was the most numerous noneconomic pest species of southeastern row crops, comprising 11.4% of total adults and nymphs of all species, and was found predominantly in wheat, with 84.2% found in wheat in 2009. The red-shouldered stink bug, Thyanta custator (F.), the dusky stink bug, Euschistus tristigmus (Say), and the spined soldier bug, Podisus maculiventris (Say), were < 1% of total captures. A range of host and nonhost weeds
4
JOURNAL OF INSECT SCIENCE
and trees bordered all farmscapes. Woods along fields comprised a number of different species; main species were loblolly pine, Pinus taeda L., water oak, Quercus nigra L., southern red oak, Quercus falcata Michaux, black cherry, Prunus serotina Ehrhart, elderberry, Sambucus canadensis L., and American sweetgum, Liquidambar styraciflua L. Other weed hosts included vetch, Vicia spp., peppergrass, Lepidium virginicum L., coffee senna, and Cassia occidentalis L. In wheat (present only in Lee County), stink bugs were first detected in early April, with E. servus sampled when flag leaves were visible (Zadoks stage 37). Numbers of adult E. servus increased rapidly as wheat entered the boot stage, with adult N. viridula first collected in wheat during heading (stage 50). Nymphs of either species were not collected in wheat in any year before the dough stage (Fig. 1A–C for E. servus). C. hilaris was not collected in wheat. The numbers of stink bugs collected decreased in wheat as the grain matured. Stink bugs were next detected in corn, with E. servus found in the V10 stage in Lee (Fig. 1A–C) and Barnwell Counties. C. hilaris was also found for the first time in Barnwell County in corn at V10 in 2009 (Fig. 2D), though this species was not found in corn in Lee County in any year, and corn was not available for sampling in Tift County. N. viridula was not found in corn in Lee County in 2009 and 2011, though 0.1 insects per 50 plants sampled were found only at V10 in 2010. In Barnwell County, where wheat was not present, adult E. servus were found in corn, with a maximum density of 1.5 insects per 50 sweeps in the milk stage, with densities decreasing as the corn matured (data not shown due to low densities). Stink bugs were first detected in cotton during squaring, and all three major species were found from stage V10 to R1 in soybean at approximately the same time in all locations where both crops were present. The first adult population peaks in cotton for E. servus and N. viridula occurred in the first WOB in all locations, with C. hilaris also peaking in the first WOB in Tift County in 2010 (data not shown). Peak populations for all three species in soybean did not occur until R4 for adults, with a sharp increase in nymphs for all species in soybean at R6, typically 3–4 wk after adult peaks (Figs. 1 and 2). Double-cropped soybean (Fig. 1A–C), planted after wheat had been harvested, also had adult peaks around R4, with a nymph stage peak at R6. C. hilaris was rarely detected in peanut (Fig. 2D), whereas densities of E. servus in peanut were also low (Fig. 2A–C). N. viridula was not found in peanut (Fig. 1D). Examining densities of adults and nymphs by species at all sampled farmscapes, E. servus showed two distinct peaks (one in wheat and one in soybean) (Fig. 1A–C). A single peak of E. servus occurred in 2011 in mid-summer in Lee County. No clear pattern could be detected for E. servus in Tift County, likely due to low densities (Fig. 2). In Tift County in 2011, where sampling was not undertaken until August, a limited number (0.02 per 50 sweeps) of nymphs were found in the first WOB in cotton, as cotton had been replanted due to severe drought. C. hilaris, found in lower numbers than E. servus, showed two adult peaks in Barnwell County in 2009, with one occurring in corn at R1 and soybean at V10, concurrently, and the other occurring at R4 in soybean, with low numbers of nymphs in both cases (Fig. 2D). Adult peaks in full-season soybean and peanut were recorded 2 wk after the application of insecticide (Fig. 2). Two peaks for C. hilaris adults were recorded at R1 and R6 and one peak for nymphs at R6 in Lee County in soybean in 2009 (data not shown). Densities were low for C. hilaris in 2010 and 2011, with only single peaks of adults and nymphs each year occurring at R6 in soybean and in the eighth WOB in cotton in Lee County. Data for N. viridula are shown only for Tift County in 2009 due to low densities and no clear patterns in all years and farmscapes (Fig. 1D). Nymphs of N. viridula increased in wheat in Lee County in 2009 at stage 87 (hard dough), following a smaller adult peak during stage 73 (early milk). A peak of adults was found in cotton in 2010, in the sixth WOB, 5 wk after the field received the only insecticide application of the year. A peak in nymphs at the first WOB in cotton was also recorded in Tift County in 2009, though densities never increased beyond 0.03 insects per 50 sweeps in cotton samples (Fig. 1D).
VOLUME 15
Spatial Patterns. Indices of dispersion are presented for farmscapes by crop and year where adequate data were available (Table 2). Populations in watermelon or sorghum were insufficient for analysis. Adults were significantly aggregated in 43 of 56 comparisons (variance to mean ratio >1, Table 2). E. servus was aggregated 39.5% of the time. Aggregated distributions for N. viridula (11.6%) and C. hilaris (9.3%) occurred less often, with the remaining 16.4% for the combined totals of adults of all species. Most aggregated distributions were found in soybean (32.6%), with cotton (20.9%), corn (18.6%), fallow (4.7%), and peanut (2.3%) showing lower percentages. Adult stink bugs had aggregated distributions in 9.3% of farmscape-date combinations in both wheat and double-cropped soybean. Lee County held 39.3% of aggregated distributions for adults, with 19.6% in Barnwell County and 16.1% in Tift County (Table 2). Nymphs were aggregated in 27 of 30 distributions (90.0%), with 33.3% for E. servus, 14.8% for N. viridula, 3.7% for C. hilaris, and the remaining 38.6% of aggregated distributions found for the combined totals of adults of all species. Nymphs were most often aggregated in soybean (48.1%), with fewer aggregated distributions in fallow (18.5%), wheat (14.8%), double-cropped soybean (11.1%), cotton (3.7%), and peanut (3.7%). The majority of significant indices for nymphs were found in Lee County (53.3%), with 6.7% in Barnwell County and 30.0% in Tift County (Table 3). Data were sufficient for analysis using Taylor’s power law and Iwao’s patchiness regression in 86 year-farmscape-crop combinations (56 for adult stink bugs and 30 for nymphs) (Tables 2 and 3). Slopes for Taylor’s power law were significantly (P < 0.05) different from one, indicating a nonrandom distribution, for adults in 22 of 56 yearfarmscape-crop combinations (39.3%) in Lee (19.6%), Barnwell (5.4%), and Tift (14.3%) counties (Table 2), whereas slopes were significantly different from one for nymphs in 10 of 30 regressions (33.3%) for Lee (26.6%), Barnwell (3.3%), and Tift (3.3%) (Table 3). For nymphs, slopes of Taylor’s power law were generally >1, indicating aggregated distributions with the exception of two in Lee County in 2009, where nymphs of all species combined in cotton and N. viridula nymphs in double-cropped soybean had uniform distributions. The density contagiousness coefficient b of Iwao’s patchiness regression was significantly different from one in 20 of 56 cases (35.7%) for adults in Lee (19.6%), Barnwell (1.8%), and Tift Counties (14.3%) (Table 2), and six of 30 cases (20.0%) for nymphs in Lee County (13.3%), with 3.3% each in Barnwell and Tift Counties (Table 3). All significant density contagiousness coefficients indicated aggregation, with the exception of one for nymphs of E. servus in wheat in Lee County in 2010, which indicated a uniform distribution. SADIE aggregation indices for year-end summary data were significant in 11 of 60 analyses (18.3%) for the three main pest species and all species summed with six (54.5%) significant indices for adults and five (45.5%) for nymphs (Table 4). Significant year-end SADIE indices indicated aggregated distributions in adults and nymphs, and all 22 significant associations between adults and nymphs were positive out of 28 paired datasets (Table 4). SADIE also was used in 608 separate sample datasets for weekly totals separated by species for all three farmscapes and years. Of those, 258 adult and 127 nymph datasets contained captures at two or more sampling locations, permitting analysis. Adults and nymphs had significant patches or gaps in 11.2% and 22.8% of analyzed weekly datasets, respectively (see weekly indices in Figs. 1 and 2). Adults and nymphs of E. servus (41.4% and 48.3%, respectively) and C. hilaris (10.3%, 6.9%) had more significant indices than N. viridula (3.4% and 3.4%). The majority of significant indices were from combined totals of adults (44.9%) and nymphs (41.4%). All significant SADIE indices indicated aggregation with the exception of uniform distributions for adult N. viridula in Lee County on 22 March 2009, E. servus nymphs in Lee County on 12 May 2009, and adult E. servus in Tift County on 22 July 2010. The SADIE association tool detected significant associations between adult and nymph stink bugs in 80.0% of 96 paired weekly datasets, with 20.8% of all significant
2015
PILKAY ET AL.: SPATIAL AND TEMPORAL DYNAMICS OF STINK BUGS
5
Fig. 1. Average densities for selected stink bug species ( 6 SEM) and daily SADIE indices of dispersion in mixed crop farmscapes in Lee County, SC (A–C, 2009–2011), and Tift County, GA (D, 2009). Arrows indicate insecticide applications. Crop phenology indicated by vegetative (V) and reproductive stages (R) in soybean and corn. Cotton stages are indicated by WOB. Wheat stages follow Zadoks scale. Peanut remained reproductive throughout the sampling periods. Asterisks indicate significant (P < 0.025) aggregations.
6
JOURNAL OF INSECT SCIENCE
VOLUME 15
Fig. 2. Average densities ( 6 SEM) and daily SADIE indices of dispersion for selected stink bug species in mixed crop farmscapes in Tift County, GA (A–C, 2009–2011), and Barnwell County, SC (D, 2009). Arrows indicate insecticide applications. Crop phenology indicated by vegetative (V) and reproductive stages (R) in soybean and corn. Cotton stages are indicated by WOB. Peanut remained reproductive throughout the sampling periods. Asterisks indicate significant (P < 0.025) aggregations. associations for E. servus, 3.1% for C. hilaris, and 4.2% for N. viridula. Associations between the combined total adults and nymphs represented 71.9% of the significant associations. All associations of adults and nymphs were positive, with the exception of a limited number in
2009, with E. servus on 19 May, N. viridula on 12 July, and all species combined on 19 May and 24 August negatively associated in Lee County, and 31 August in Tift County, where E. servus adults and nymphs were negatively associated. The limited number of negative
2015
PILKAY ET AL.: SPATIAL AND TEMPORAL DYNAMICS OF STINK BUGS
7
Table 2. Dispersion indices for stink bug adults in selected crops from Lee and Barnwell Counties, SC, and Tift County, GA Location
Year
Crop
Species
Taylor’s power law a
b
R2
Iwao’s regression
t-value for slope ¼ 1
a
b
R2
ID
t-value for slope ¼ 1
Barnwell 2009 Corn E. servus 1.386 1.239 0.98 1.759 0.061 1.448 0.87 1.363 1.371a Barnwell 2009 Corn All 1.163 1.001 1.00 0.018 0.020 1.185 0.93 0.988 1.487a Barnwell 2009 Cotton E. servus 1.566 0.681 0.71 0.797 0.351 3.026 0.19 0.578 2.267a Barnwell 2009 Cotton All 1.566 0.681 0.71 0.797 0.351 3.026 0.19 0.578 2.267a Barnwell 2009 Soybean E. servus 0.771 0.773 0.76 0.757 0.089 1.000 0.09 0 0.917 Barnwell 2009 Soybean C. hilaris 92.073 4.501 0.67 1.874 0.360 8.596 0.37 1.791 3.159a Barnwell 2009 Soybean All 4.724 1.773 0.62 0.932 0.991 6.327 0.50 2.245 2.842a Barnwell 2010 Cotton E. servus 3.506 1.491 0.98 3.186* 0.162 3.940 0.59 1.554 1.154a Barnwell 2010 Cotton All 1.672 1.126 0.98 1.297 0.004 2.958 0.65 2.831* 1.346a Barnwell 2010 Soybean All 0.954 0.557 0.69 1.125 0.464 0.608 0.05 0.624 1.247 Barnwell 2011 Corn E. servus 1.251 3.681 0.81 3.023* 0.406 2.677 0.67 2.173 2.083a Barnwell 2011 Corn All 1.251 3.681 0.81 3.023* 0.406 2.677 0.67 2.173 2.084a Barnwell 2011 Cotton E. servus 7.092 1.926 0.98 2.555 0.063 2.670 0.58 1.353 1.210a Barnwell 2011 Cotton All 3.080 1.479 0.97 2.174 0.156 3.271 0.59 1.818 1.123 Lee 2009 Cotton E. servus 1.141 1.108 1.00 4.750* 0.051 1.174 0.98 2.155 1.429a Lee 2009 Cotton All 1.122 1.319 0.98 4.132* 0.117 1.044 0.75 0.216 1.001 Lee 2009 Soybean E. servus 2.248 1.322 0.56 0.75 0.259 2.740 0.48 2.11 1.936a Lee 2009 Soybean N. viridula 1.947 1.101 0.78 0.25 0.245 4.931 0.30 1.877 1.639a Lee 2009 Soybean C. hilaris 0.720 0.677 0.83 1.707 0.103 1.121 0.17 0.17 1.216a Lee 2009 Soybean All 2.379 1.750 0.66 1.597 0.317 2.663 0.52 2.266* 2.354a Lee 2009 Wheat E. servus 1.927 2.010 1.00 14.064* 0.455 2.680 0.96 8.671* 5.099a Lee 2009 Wheat N. viridula 2.301 0.981 0.63 0.035 0.255 6.503 0.23 1.552 2.364a Lee 2009 Wheat All 1.530 2.179 0.99 7.371* 0.226 2.457 0.88 4.730* 4.869a Lee 2009 DSB E. servus 7.467 2.241 0.82 2.057 0.326 3.677 0.49 2.841* 1.303a Lee 2009 DSB N. viridula 1.756 0.987 0.99 0.182 0.011 2.507 0.83 3.184* 2.017a Lee 2009 DSB C. hilaris 7.151 1.708 0.98 4.252* 0.413 9.266 0.91 9.611* 1.899a Lee 2009 DSB All 1.913 1.131 0.92 1.028 0.104 2.291 0.76 3.518* 1.908a Lee 2010 Corn E. servus 26.174 3.863 0.96 4.335* 0.135 2.118 0.63 1.934 1.381a Lee 2010 Corn All 12.701 3.230 0.99 10.216* 0.321 3.401 0.66 2.672* 1.767a Lee 2010 Soybean E. servus 0.980 0.719 0.40 0.648 0.035 1.392 0.16 0.44 1.265a Lee 2010 Soybean C. hilaris 1.846 1.318 1.00 6.094* 0.265 1.534 0.91 1.341 2.246a Lee 2010 Soybean All 2.022 0.776 0.60 0.828 0.205 1.790 0.42 1.241 2.222a Lee 2010 Wheat All 2.255 2.955 0.97 6.498* 1.073 3.693 0.74 4.005* 2.224a Lee 2011 Corn E. servus 0.016 7.449 1.00 18.622* 0.561 5.293 0.86 4.158* 12.201a Lee 2011 Corn All 0.016 7.449 1.00 18.622* 0.561 5.293 0.86 4.158* 12.201a Lee 2011 Cotton E. servus 1.605 1.183 0.90 0.571 0.095 2.485 0.17 1.021 1.113 Lee 2011 Cotton All 1.219 0.960 0.63 0.088 0.038 3.363 0.09 0.756 1.314a Lee 2011 Soybean E. servus 1.545 0.779 0.67 0.605 0.497 1.121 0.16 0.179 1.904a Lee 2011 Soybean All 1.649 0.645 0.76 1.776 0.538 1.113 0.30 0.224 2.117a Tift 2009 Cotton E. servus 1.126 1.066 0.98 0.972 0.050 1.233 0.54 0.603 1.047 Tift 2009 Cotton All 0.828 0.900 0.94 0.901 0.073 0.629 0.03 0.331 1.074 Tift 2009 Fallow E. servus 1.760 1.226 0.99 3.167 0.086 1.668 0.64 1.299 1.794a Tift 2009 Fallow All 2.004 1.223 1.00 3.783* 0.132 1.923 0.72 1.906 1.988a Tift 2009 Soybean E. servus 3.223 1.239 0.67 0.386 0.492 5.717 0.39 1.845 2.548a Tift 2009 Soybean N. viridula 1.708 0.994 0.77 0.014 0.175 4.135 0.22 1.229 1.753a Tift 2009 Soybean All 2.311 1.002 0.86 0.01 0.026 2.992 0.57 2.204* 2.482a Tift 2010 Cotton E. servus 1.169 1.040 0.96 0.328 0.040 1.984 0.38 1.133 1.087 Tift 2010 Cotton N. viridula 0.593 0.775 0.87 1.348 0.170 1.510 0.01 0.068 1.211a Tift 2010 Cotton C. hilaris 1.056 1.008 0.95 0.06 0.026 1.671 0.17 0.53 1.062 Tift 2010 Cotton All 1.477 1.063 0.98 0.81 0.032 2.427 0.72 2.676* 1.399a Tift 2010 Fallow E. servus 3.543 1.451 1.00 4.676* 0.226 5.623 0.92 4.656* 1.240a Tift 2010 Fallow All 4.262 1.553 1.00 6.598* 0.326 6.163 0.90 4.579* 1.196 Tift 2010 Peanut E. servus 2.331 1.438 1.00 10.985* 0.214 2.783 0.93 6.306* 1.398a Tift 2010 Peanut All 2.293 1.418 1.00 10.377* 0.195 2.733 0.93 6.480* 1.359 Tift 2011 Fallow E. servus 2.212 1.374 0.99 4.220* 1.145 1.333 0.59 3.650* 1.099 Tift 2011 Fallow All 2.212 1.374 0.99 4.220* 1.145 1.333 0.59 3.650* 1.069 Locations and crops with insufficient samples for analysis have been omitted. Double-crop soybeans are indicated by “DSB.” ID, overall index of dispersion, aggregated ( > 1), random (1), or uniform ( < 1). a 2 v test indicated significant difference from 1 (P < 0.05; in bold). *P < 0.05 (in bold).
associations suggested significant clusters of adults and nymphs were generally found in the same area of the farmscape. IDW interpolation maps for weekly SADIE aggregation indices are presented for E. servus adults and nymphs in Lee County in 2009 (Fig. 3). In Lee County in 2009, E. servus nymphs were aggregated in four weeks of sampling, as opposed to three for adults (Fig. 3). Clusters of adults were located in wheat on 6/9, in double cropped soybean, cotton, and full season soybeans on 7/21, and in full season soybeans on 8/4. Clusters of nymphs were located in wheat on 5/26 and 6/9, and
in full season soybeans on 8/24 and 9/1. Peak populations in corn were rarely above 0.5 stink bugs per 50 plants, and clustering was not observed. Soybean and cotton adjacent to one another demonstrated adult clustering in late July (Figs. 1 and 3). In Barnwell County in 2009, C. hilaris demonstrated significant adult clustering in fields of cotton and soybean adjacent to one another on 1 July 2009. Distance From Field Borders and Landscape Effects. Distance from field edges did not have a significant effect on stink bug densities in any farmscape sampled (P > 0.05) (Table 5). However, crop and adjacent
8
JOURNAL OF INSECT SCIENCE
VOLUME 15
Table 3. Dispersion indices for stink bug nymphs in selected crops from Lee and Barnwell Counties, SC, and Tift County, GA Location
Year
Crop
Species
Taylor’s power law a
b
R2
Iwao’s regression
t-value for slope ¼ 1
a
b
R2
ID
t-value for slope ¼ 1
Barnwell 2009 Soybean All 1.499 1.044 0.93 0.282 0.030 3.424 0.34 1.353 1.339a Barnwell 2010 Soybean All 3.734 2.391 1.00 8.652* 1.733 5.533 0.94 7.519* 3.489a Lee 2009 Cotton All 0.615 0.326 0.51 2.319* 0.540 0.294 0.02 0.749 1.478 Lee 2009 Soybean E. servus 4.799 1.064 0.60 0.091 0.293 3.076 0.41 1.915 4.631a Lee 2009 Soybean N. viridula 3.958 0.780 0.93 1.742 2.746 1.168 0.11 0.107 4.823a Lee 2009 Soybean C. hilaris 2.174 0.851 0.70 0.389 0.174 3.988 0.22 1.154 2.691a Lee 2009 Soybean All 12.925 0.482 0.68 1.762 3.906 1.966 0.25 0.835 8.437a Lee 2009 Wheat E. servus 2.653 1.260 0.99 3.013 0.976 1.467 0.82 1.405 5.646a Lee 2009 Wheat N. viridula 4.163 0.868 0.99 2.431* 0.452 2.311 0.78 1.887 5.218a Lee 2009 Wheat All 3.367 1.005 1.00 0.181 1.313 1.218 0.91 1.187 7.118a Lee 2009 DSB E. servus 1.840 8.317 1.00 10.498* 0.358 2.468 0.96 5.163* 2.781a Lee 2009 DSB N. viridula 3.066 1.495 1.00 6.972* 0.088 2.591 0.94 4.612* 5.135a Lee 2009 DSB All 1.171 1.945 1.00 11.779* 0.123 1.741 0.94 3.707* 5.433a Lee 2010 Soybean E. servus 0.000 21.453 0.99 8.174* 0.223 2.735 0.64 1.955 5.274a Lee 2010 Soybean N. viridula 4.755 1.554 0.98 2.829* 0.399 3.946 0.35 1.837 2.073a Lee 2010 Soybean All 4.670 0.906 0.62 0.139 0.096 2.336 0.61 1.669 5.709a Lee 2010 Wheat E. servus 1.399 0.000 0.24 0.005 6.684 2.863 0.41 2.462* 2.18 Lee 2010 Wheat All 0.757 1.649 0.96 2.948* 0.196 1.156 0.90 0.924 2.179a Lee 2011 Soybean E. servus 12.767 0.733 0.46 0.354 1.020 1.989 0.46 1.158 11.413a Lee 2011 Soybean All 20.965 0.452 0.56 1.863 3.700 1.233 0.47 0.409 13.871a Tift 2009 Fallow E. servus 2.439 0.944 0.96 0.448 0.980 2.359 0.20 0.586 2.714a Tift 2009 Fallow All 2.498 1.024 0.95 0.149 0.560 3.000 0.29 0.926 2.520a Tift 2009 Soybean E. servus 2.786 1.427 0.93 0.613 0.230 3.530 0.74 2.29 2.324a Tift 2009 Soybean All 1.882 0.614 0.89 1.833 0.762 2.411 0.17 0.742 2.691a Tift 2010 Cotton All 240.718 2.882 1.00 10.926* 1.047 34.257 0.84 5.273* 2.524a Tift 2010 Fallow E. servus 2.223 1.476 0.83 0.728 0.092 2.081 0.46 0.831 1.240a Tift 2010 Fallow All 2.647 1.694 0.91 1.196 0.069 1.822 0.66 1.082 1.215a Tift 2010 Peanut All 1.341 0.885 0.99 1.939 0.438 1.150 0.24 0.11 1.755a Tift 2011 Fallow E. servus 1.558 1.092 0.95 0.627 0.032 2.558 0.55 1.335 1.333a Tift 2011 Fallow All 1.516 1.133 0.95 0.769 0.051 2.143 0.60 1.301 1.254 Locations and crops with insufficient samples for calculation have been omitted. Double-crop soybeans are indicated by “DSB.” ID, overall index of dispersion, aggregated ( > 1), random (1), or uniform ( < 1). a 2 v test indicated significant difference from 1 (P < 0.05; in bold). *P < 0.05 (in bold).
crop effects were often significant and contrasts were developed to compare insect densities in sets of similar crop and adjacent crop combinations with densities in other sets of crop and adjacent crop combinations. For example, in Lee County, crop and adjacent crop effects were significant (P < 0.05) and densities in crop and adjacent crop combinations with full-season soybean or woods (consisting of cotton fields adjacent to soybean fields, cotton fields adjacent to woods, soybean fields adjacent to cotton fields, wheat fields adjacent to woods, and soybean fields adjacent to woods) were significantly higher than densities in the other combinations (i.e., corn and cotton, corn and wheat-double-crop soybean, corn and woods, wheat and cotton). Specifically, average stink bug density was higher by 0.20 6 0.07 (SEM) (t ¼ 2.74; df ¼ 13.2; P ¼ 0.0167), adults of E. servus, N. viridula were higher by 0.11 6 0.03 (t ¼ 3.64; df ¼ 11.78; P ¼ 0.0035), and combined adults of all species were higher by 0.46 6 .0.11 (t ¼ 4.04; df ¼ 13.83; P ¼ 0.0012). Densities of nymphs of E. servus were higher by 1.04 6 0.28 (t ¼ 3.70; df ¼ 12.11; P ¼ 0.0030) in soybean versus cotton fields. This also occurred for nymphs of C. hilaris (0.57 6 0.08; t ¼ 7.14; df ¼ 12.76; P < 0.0001) and combined nymphs of all species (1.83 6 0.31; t ¼ 5.88; df ¼ 12.38; P < 0.0001). Adults of C. hilaris and nymphs of N. viridula were not significantly influenced by crop and adjacent crop effects (P > 0.05). Barnwell County, lacking woods transects, still exhibited higher estimated densities on soybean fields adjacent to cotton or peanut fields and cotton fields adjacent to soybean or peanut fields for C. hilaris adults (0.31 6 0.03; t ¼ 10.53; df ¼ 41; P < 0.0001), as well as adults (0.46 6 0.08; t ¼ 6.09; df ¼ 25.66; P < 0.0001) and nymphs (0.03 6 0.01; t ¼ 2.98; df ¼ 164; P < 0.0033) of all species combined. In Tift County, no crop and adjacent crop effect of any crop combination influenced stink bug densities for any species. Interactions between crop and adjacent crop effects and distance were
detected in Lee County for E. servus adults, adults and nymphs of N. viridula, nymphs of C. hilaris, and the combined nymphs of all species (Table 5). However, post-hoc analyses using contrast statements and mean separation tests did not reveal any biologically meaningful trends.
Discussion Although SADIE and IDW of local aggregation indices have been used previously to describe the spatial dynamics of stink bugs (Tillman et al. 2009, Reay-Jones et al. 2010), this study is the first to attempt to use these techniques to quantify the spatial and temporal dispersal of adults and nymphs of multiple species across multiple years and farmscapes in different states. SADIE detected fewer aggregations than the variance-to-mean ratio, Taylor’s power law, or Iwao’s patchiness regression, with 82% of SADIE analyses indicating randomness. Slopes of Taylor’s power law were greater than 1 in 31% of analyses, indicating a clumped distribution for adult and nymph stink bugs. Fit of Iwao’s patchiness regression also generally indicated clumped distributions for adult and nymph stink bugs (b > 1) when b was significantly different from one, but distributions were random in 64% of analyses, supporting the results of Taylor’s power law. The majority of significant slopes for both regressions were found in areas of highest densities, such as in Lee County or in soybean. Many arthropod species, including stink bugs, are spatially aggregated in fields of crops (Taylor et al. 1978; Reay-Jones et al. 2009, 2010; Reay-Jones 2012). The degree of aggregation can vary with species and life stage. Based on SADIE, nymphs were slightly more frequently aggregated than adults. Stink bug eggs are laid in masses, and nymphs do not disperse from ovipositional sites until maturation to late instars (Kiritani et al. 1965), as illustrated by the high counts in soybean in Lee County, with up to 16 nymphs in a single sample. However,
2015
PILKAY ET AL.: SPATIAL AND TEMPORAL DYNAMICS OF STINK BUGS
Table 4. SADIE summary data analyses for year-end total stink bug dispersion indices across all crops by location for 2009–2011 Location Year
Barnwell Barnwell Barnwell Barnwell Barnwell Barnwell Barnwell Barnwell Barnwell Barnwell Barnwell Barnwell Lee Lee Lee Lee Lee Lee Lee Lee Lee Lee Lee Lee Tift Tift Tift Tift Tift Tift Tift Tift Tift Tift Tift Tift
2009 2009 2009 2009 2010 2010 2010 2010 2011 2011 2011 2011 2009 2009 2009 2009 2010 2010 2010 2010 2011 2011 2011 2011 2009 2009 2009 2009 2010 2010 2010 2010 2011 2011 2011 2011
Species
E. servus C. hilaris N. viridula All E. servus C. hilaris N. viridula All E. servus C. hilaris N. viridula All E. servus C. hilaris N. viridula All E. servus C. hilaris N. viridula All E. servus C. hilaris N. viridula All E. servus C. hilaris N. viridula All E. servus C. hilaris N. viridula All E. servus C. hilaris N. viridula All
Adult
Nymph
Ia
Pa
Ia
2.022* 1.195 1.058 1.825* 1.175 1.180 — 1.309 0.987 0.887 — 1.054 1.249 0.885 1.310 1.361 1.289 1.667* 1.024 1.389 1.852* 0.813 — 1.835* 1.292 0.853 1.524* 1.441 1.119 1.124 1.008 1.207 1.049 — — 1.103
0.0002 0.1421 0.3204 0.0007 0.1584 0.1537 — 0.0754 0.4597 0.6906 — 0.3377 0.0984 0.7178 0.0603 0.0385 0.0670 0.0017 0.3752 0.0275 0.0003 0.8966 — 0.0003 0.0771 0.7783 0.0144 0.0258 0.2226 0.2041 0.4153 0.1354 0.3270 — — 0.2422
— 1.037 1.418 1.156 1.116 1.023 1.118 0.958 — 1.312 — 1.167 1.427 1.103 1.357 1.472* 2.016* 1.477* 1.151 2.058* 1.041 1.115 — 1.070 1.259 — 1.368 1.435 1.030 0.899 1.139 1.104 1.499* — — 1.425
Pa
Association X
— — 0.3476 0.464† 0.0273 0.367† 0.1820 0.598† 0.2247 0.485† 0.3866 0.326† 0.2306 — 0.5247 0.324† — — 0.0618 0.434† — — 0.1798 0.090 0.0255 0.569† 0.2423 0.293 0.0402 0.535† 0.0146 0.706† 0.0002 0.579† 0.0132 0.491† 0.1766 0.439† 0.0002 0.617† 0.3472 0.360† 0.2199 0.413† — — 0.2961 0.384† 0.0897 0.291† — — 0.0474 0.582† 0.0282 0.398† 0.3617 0.118 0.6524 0.028 0.1966 0.123 0.2517 0.165 0.0188 0.359† — — — — 0.0288 0.407†
P(X) — 0.0089 0.0151
